created script to identify decoy genes (previously referred to as sus…

…picious genes) for each cancer subtype

scripts/identify_decoy_genes.py

@@ -0,0 +1,58 @@
+import os
+import logging
+import pandas as pd
+
+from argparse import ArgumentParser
+
+
+# ---------------------------------------------------------------------------- #
+#                               HELPER FUNCTIONS                               #
+# ---------------------------------------------------------------------------- #
+def build_argument_parser():
+    parser = ArgumentParser(description="Identify decoy genes")
+    parser.add_argument(
+        "-c", "--cnt", required=True, help="Path to the count matrix file"
+    )
+    parser.add_argument(
+        "-d", "--drvr", required=True, help="Path to the driver genes file"
+    )
+    parser.add_argument(
+        "-k",
+        "--top_k",
+        type=int,
+        default=100,
+        help="Rank cutoff for high mutation frequency",
+    )
+    parser.add_argument("-s", "--subtype", required=True, help="Name of cancer subtype")
+    parser.add_argument(
+        "-o", "--out", required=True, help="Path to the output directory"
+    )
+    return parser
+
+
+# ---------------------------------------------------------------------------- #
+#                                MAIN FUNCTIONS                                #
+# ---------------------------------------------------------------------------- #
+def identify_decoy_genes(cnt_df, driver_genes, k, fout):
+    logging.info("Identifying decoy genes")
+    gene_mutation_counts = cnt_df.sum(axis=0).sort_values(ascending=False)
+    top_k_genes = gene_mutation_counts.head(k).index
+    decoy_genes = [gene for gene in top_k_genes if gene not in driver_genes]
+    with open(fout, "w") as f:
+        f.write("\n".join(decoy_genes))
+
+
+if __name__ == "__main__":
+    parser = build_argument_parser()
+    args = parser.parse_args()
+    cnt_df = pd.read_csv(args.cnt, index_col=0)
+    drvr_df = pd.read_csv(args.drvr, sep="\t", index_col=0)
+    driver_genes = set(drvr_df.index + "_M") | set(drvr_df.index + "_N")
+    os.makedirs(args.out, exist_ok=True)
+    fout = os.path.join(args.out, f"{args.subtype}_decoy_genes.txt")
+    identify_decoy_genes(
+        cnt_df=cnt_df,
+        driver_genes=driver_genes,
+        k=args.top_k,
+        fout=fout,
+    )